VISION LAB  
Efficient and Exact Inference using BranchandBound
We propose a novel BranchandBound (BB) method to efficiently solve exact MAPMRF inference on problems with a large number of states (per variable). These problems are common in computer vision or computational biology. For example, in computer vision, the problem of estimating the pose of a person (i.e., body part configurations, see figure) can be addressed by modeling the human body using a MRF wherein states correspond to locations and orientations of each body part. In computational biology, protein design can be modeled using a MRF model wherein states corresponding to combinations of different amino acids and rotamers. Our BB approach leverages a novel data structure to reduce the time complexity and a novel branching strategy that reduces the number of branches that are to be evaluated. Extensive theoretical and experimental analysis demonstrates that our method is provably faster than naive BB algorithms and other stateoftheart methods [3,4] for extract inference on loopy believe models. Anecdotal results are shown in the figure on the right. Publications
Source Codes
References
Contact : sunmin at umich dot edu Last update : April 3rd, 2012
